@

초록

Chronic conditions-such as cardiovascular disease, cancer, and diabetes mellitus-represent a significant global healthcare burden due to the requirement for lifelong treatment and management. These conditions necessitate substantial healthcare expenditure, making early detection and prevention critical for improving patient outcomes and lowering overall costs. Artificial intelligence (AI), especially machine learning (ML), offers promising capabilities for pattern recognition and data analysis within chronic disease datasets, thereby enhancing clinical understanding and decision-making. This study addresses both the challenges and opportunities in chronic disease prediction, emphasizing the importance of collaborative efforts among healthcare professionals, data scientists, and policymakers. We propose a novel five-layered ML framework integrated with a decision support system (DSS) to facilitate the early identification of chronic cardiac disorders. In this framework, hyperparameters were systematically fine-tuned for five sion, and eXtreme Gradient Boosting (XGBoost). To validate our model, we utilized real-world datasets on chronic cardiac disease, assessing its predictive performance through rigorous simulations. Our proposed ML-based DSS achieved an accuracy of 94.27% utilizing all available features, and 95.19% when applying a feature selection strategy, underscoring its effectiveness for chronic disease prediction and clinical utility.

키워드